"Completely Automated Public Turing test to tell Computers and Humans Apart",
Gosh!
This has sparked a random thought: If answering a mathematics question is used as this test then we have a couple of tricky issues to resolve.
(1) a CAS is often much more accurate than a person with a complex "symbol-pushing" calculation.
So, what happens if the computer incorporates a CAS as an attempt to fool this test?
(2) people are nororiously poor at basic logical reasoning. Have a look at the "Wasson Selection test" as a prime example.
http://en.wikipedia.org/wiki/Wason_selection_task
and with specific reference to mathematicians see the work of Matthew Inglis.
http://www.lsri.nottingham.ac.uk/mji/index.htm
I'm not trying to criticise this approach to reducing SPAM - I hope it works. It does raise interesting questions in my mind about what we are trying to achieve when we ask students similar questions as part of our assessments.....
Chris
Trackback URL for this post:
- Ajouter un commentaire
- 1894 lectures



Chris, of course, a (not so)
Chris, of course, a (not so) smart spammer will be able to easily fool this. In fact, there are already attempts at cracking captcha's, but so far so good. Before we used it on the registration page, we used to get tens of fake registrations each day. Now we get none despite the triviality of the question. In fact, just asking to retype a string would also do it mainly because a generic spammer does not go through the trouble of trying to beat these tests.
My reasoning for choosing this trivial mathematical test, as apposed to the image-based ones, was that
And I do not think you ask your students to do sums of 1-digit numbers but you do assume that they know how to do it.